Combining generative and discriminative models for semantic segmentation of CT scans via active learning

Juan Eugenio Iglesias, Ender Konukoglu, Albert Montillo, Zhuowen Tu, Antonio Criminisi

Research output: Chapter in Book/Report/Conference proceedingConference contribution

32 Scopus citations

Abstract

This paper presents a new supervised learning framework for the efficient recognition and segmentation of anatomical structures in 3D computed tomography (CT), with as little training data as possible. Training supervised classifiers to recognize organs within CT scans requires a large number of manually delineated exemplar 3D images, which are very expensive to obtain. In this study, we borrow ideas from the field of active learning to optimally select a minimum subset of such images that yields accurate anatomy segmentation. The main contribution of this work is in designing a combined generative-discriminative model which: i) drives optimal selection of training data; and ii) increases segmentation accuracy. The optimal training set is constructed by finding unlabeled scans which maximize the disagreement between our two complementary probabilistic models, as measured by a modified version of the Jensen-Shannon divergence. Our algorithm is assessed on a database of 196 labeled clinical CT scans with high variability in resolution, anatomy, pathologies, etc. Quantitative evaluation shows that, compared with randomly selecting the scans to annotate, our method decreases the number of training images by up to 45%. Moreover, our generative model of body shape substantially increases segmentation accuracy when compared to either using the discriminative model alone or a generic smoothness prior (e.g. via a Markov Random Field).

Original languageEnglish (US)
Title of host publicationInformation Processing in Medical Imaging - 22nd International Conference, IPMI 2011, Proceedings
PublisherSpringer Verlag
Pages25-36
Number of pages12
Volume6801 LNCS
ISBN (Print)9783642220913
DOIs
Publication statusPublished - Jan 1 2011
Event22nd International Conference on Information Processing in Medical Imaging, IPMI 2011 - Kloster Irsee, Germany
Duration: Jul 3 2011Jul 8 2011

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume6801 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other22nd International Conference on Information Processing in Medical Imaging, IPMI 2011
CountryGermany
CityKloster Irsee
Period7/3/117/8/11

    Fingerprint

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Iglesias, J. E., Konukoglu, E., Montillo, A., Tu, Z., & Criminisi, A. (2011). Combining generative and discriminative models for semantic segmentation of CT scans via active learning. In Information Processing in Medical Imaging - 22nd International Conference, IPMI 2011, Proceedings (Vol. 6801 LNCS, pp. 25-36). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6801 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-642-22092-0_3